miRNA Fetal Tissues: Data Exploration

Data Exploration

Read the data in using readxl::read_xlsx. What other ways can we read this data in?
Data can be read in depending upon the file type. So, for an excel file, we can use -
read.excel
read.xls
read.table

---

What type of data is in the magda_cleanedmiRNAdata.xlsx file? How is it structured?
The data shows micro RNA levels detected from different tissues obtained at different gestational ages of several differnt foetuses.

---

How many “variables” are present?
There are 27 variables present, namely -
Reading data in

library(dplyr)
library(tidyr)
library(readxl)
library(tidyverse)
library(DT)
library(knitr)
library(kableExtra)
library(ggplot2)
library(rmarkdown)
raw_miRNA <- read_excel("Z:/Nikita/Projects/mirna_fetal_tissues/data/external/magda_cleanedmiRNAdata.xlsx", col_names = FALSE, sheet = "miRNA")

Variables present:

print(raw_miRNA[2:28, 1], n = 27)

## # A tibble: 27 x 1
##    ...1                   
##    <chr>                  
##  1 lamID                  
##  2 gscID                  
##  3 caseID                 
##  4 tissue                 
##  5 condition              
##  6 conditionEx            
##  7 GA                     
##  8 trimester              
##  9 sex                    
## 10 processingTime         
## 11 libraryID              
## 12 flowCell               
## 13 Lane                   
## 14 indexAdapter           
## 15 MLPA                   
## 16 misoprostol            
## 17 modeLabour             
## 18 medicalTA              
## 19 totalReadsAligned_miRNA
## 20 propFullyIn_miRNA      
## 21 propPartiallyIn_miRNA  
## 22 totalReadsAligned_piRNA
## 23 propFullyIn_piRNA      
## 24 propPartiallyIn_piRNA  
## 25 RQS                    
## 26 Degraded               
## 27 450k.array

---

How many patients are present?
There are a total of 106 patients present, of which

45 are female:

raw_miRNA %>% 
  str_extract_all('FEMALE') %>%
  unlist %>%
  length
## [1] 45

61 are male:

raw_miRNA %>% 
  str_extract_all('MALE') %>%
  unlist %>%
  length
## [1] 106

Hence,
106 (word: MALE) - 45 (word: FEMALE)
= 61

---

How many samples are present?
- There are 106 total patients.
- Number of miRNAs =
2822 (total rows) - 29 (variables plus initial empty rows)
= 2793
Hence,
2793 * 106
= 296,058

---

What is the breakdown of tissues / trimester? Generate a pretty table using kableExtra::kable(). Do this with at least 1 other variable of your choosing.
I decided to subset 5 variables -
- tissue
- extended sample condition information
- gestational age
- trimester
- sex

sub_miRNa <- raw_miRNA[c(5,7,8,9,10),]
t_sub_miRNA <- t(sub_miRNa)

---

Summary Characteristics

Summary of the variables:

summary(t_sub_miRNA)

##              tissue                 conditionEx                  GA    
##  brain          :10   spina bifida        :27   22                :12  
##  chorionic villi:32   pPROM               :22   17                : 7  
##  kidney         :10   control             :18   18                : 7  
##  liver          :10   anencephaly         :10   21.6              : 7  
##  lung           :24   GU abnormalities    : 7   22.4              : 7  
##  muscle         :10   Lipomyelomeningocele: 5   20.399999999999999: 6  
##  spinal cord    :10   (Other)             :17   (Other)           :60  
##  trimester     sex    
##  1: 6      FEMALE:45  
##  2:90      MALE  :61  
##  3:10                 
##                       
##                       
##                       
## 

---

Since the data was all in a matrix form before, I converted it into a dataframe raw_miRNA.df:

t_raw_mirna <- t(raw_miRNA)
colnames(t_raw_mirna) = t_raw_mirna[1, ]
t_raw_mirna <- t_raw_mirna[-1, ]
raw_miRNA.df <- as.data.frame(t_raw_mirna)

raw_miRNA.df is now: 1. the transposed version of raw_miRNA
2. a dataframe
3. has column names
4. easier to view and understand

---

Moving all of the non-miRNA data (characteristics) into a separate df:

pDat <- raw_miRNA.df[ c(1:28) ]
colnames(pDat)[1] <- "ID"

pDat <- head(pDat,-8)

pDat <- as_tibble(pDat)
paged_table(pDat)

And only the miRNA expression level data into o_miRNA.df

o_miRNA.df <- raw_miRNA.df [-c(2:28)]

paged_table(o_miRNA.df)

---

Tissue versus Trimester

For calculating the breakdown of tissues / trimester:

tri_tis <- (pDat [(c(5,9))])

tri_tis <- rename(count(tri_tis, trimester, tissue), Freq = n) #getting frequency (sum) of all total number of samples

tri_tis <- tri_tis[, c(2,1,3)]

tri_tis <- tri_tis %>% 
            spread(trimester, Freq) %>% 
            replace(is.na(.), 0)

names(tri_tis) <- c("Tissue","Trimester_1", "Trimester_2", "Trimester_3")

#tri_tis %>%
#  mutate (Tissue = cell_spec(Tissue, color = "pink", background ="black")) %>% 
  kable(tri_tis, align = "c") %>% 
    kable_styling(bootstrap_options = c("striped", position = "centre", font_size = 12))

Tissue	Trimester_1	Trimester_2	Trimester_3
brain	0	10	0
chorionic villi	6	16	10
kidney	0	10	0
liver	0	10	0
lung	0	24	0
muscle	0	10	0
spinal cord	0	10	0

All the tissues except for chorionic villi contained samples from only trimester 2 pregnancies.

Bar Plot

Showing the above as a bar plot:

  ggplot (data = pDat, aes(x = trimester, fill=tissue)) + #max no of samples without specified y-axis legth
   labs(title="Tissues per Trimester", subtitle = "Colors denote count of tissue", x ="Trimester", y ="No. of samples") +
  # scale_fill_brewer(palette=11) + #to add colours from a shaded pallete, only works for multiple variables
   geom_bar (stat = "count", width=0.8) +
   coord_cartesian(ylim=c(0,100))